For me, the suicide-related data is so hard to measure and so open for debates, that I’d treat it separately, or not include it at all, if using death count as an argument against llms, since it’s a breach for deviating the debate.